The accuracy of two different methods in measuring the ammonia nitrogen (N-NH3) concentration in rumen fluid were evaluated: a catalyzed indophenol colorimetric reaction (CICR) and the Kjeldahl distillation (KD). Five buffered standard solutions containing volatile fatty acids, true protein, and known ammonia concentrations (0, 3, 6, 12, and 24 N-NH3 mg/dL) were used to simulate rumen fluid. Different ratios (10:1, 7.5:1, 5:1, 2.5:1, 1:1, 1:2.5, 1:5, 1:7.5, and 1:10) of a potassium hydroxide solution (KOH, 2 mol/L) to standard solutions were evaluated by the KD method. The accuracy of each method was evaluated by adjusting a simple linear regression model of the estimated N-NH3 concentrations on the N-NH3 concentrations in the standard solutions. When the KD method was used, N-NH3 was observed to be released from the deamination of true protein (P<0.05), and an incomplete recovery of N-NH3 was observed (P<0.05), except for 7.5:1 and 5:1 ratios of KOH solution to standard solutions (P>0.05). The estimates of the N-NH3 concentration obtained by the CICR method were found to be accurate (P>0.05). After the accuracy evaluation, ninety-three samples of rumen fluid were evaluated by the CICR and KD methods (using the 5:1 ratio of KOH solution to rumen fluid sample), assuming that the CICR estimates would be accurate. The N-NH3 concentrations obtained by the two methods were observed to be different (P<0.05) but strongly correlated (r = 0.9701). Thus, it was concluded that the estimates obtained by the Kjeldahl distillation using a 5:1 ratio of KOH solution to rumen fluid sample can be adjusted to avoid biases. Furthermore, a model to adjust the N-NH3 concentration is suggested.